On-Device Spatial Attention based Sequence Learning Approach for Scene Text Script Identification

TL;DR

This paper introduces a lightweight, real-time on-device CNN-LSTM model with spatial attention for scene text script identification, suitable for resource-constrained mobile devices, achieving competitive accuracy and efficiency.

Contribution

The paper presents a novel, efficient CNN-LSTM architecture with spatial attention and residual blocks for script identification, optimized for on-device deployment.

Findings

Achieves competitive accuracy with state-of-the-art methods.

Has a small model size of 1.1 million parameters.

Inference time is only 2.7 milliseconds.

Abstract

Automatic identification of script is an essential component of a multilingual OCR engine. In this paper, we present an efficient, lightweight, real-time and on-device spatial attention based CNN-LSTM network for scene text script identification, feasible for deployment on resource constrained mobile devices. Our network consists of a CNN, equipped with a spatial attention module which helps reduce the spatial distortions present in natural images. This allows the feature extractor to generate rich image representations while ignoring the deformities and thereby, enhancing the performance of this fine grained classification task. The network also employs residue convolutional blocks to build a deep network to focus on the discriminative features of a script. The CNN learns the text feature representation by identifying each character as belonging to a particular script and the long term spatial dependencies within the text are captured using the sequence learning capabilities of the LSTM layers. Combining the spatial attention mechanism with the residue convolutional blocks, we are able to enhance the performance of the baseline CNN to build an end-to-end trainable network for script identification. The experimental results on several standard benchmarks demonstrate the effectiveness of our method. The network achieves competitive accuracy with state-of-the-art methods and is superior in terms of network size, with a total of just 1.1 million parameters and inference time of 2.7 milliseconds.